You are on page 1of 11

Applied Ocean Research 35 (2012) 1424

Contents lists available at SciVerse ScienceDirect

Applied Ocean Research


journal homepage: www.elsevier.com/locate/apor

Semi-empirical methods for determining the efux velocity from


a ships propeller
W. Lam a,b, , G.A. Hamill b,1 , D.J. Robinson b,1 , S. Raghunathan b,1
a
b

Department of Civil Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia
Queens University Belfast, Northern Ireland BT9 5AG, United Kingdom

a r t i c l e

i n f o

Article history:
Received 22 October 2010
Received in revised form 3 January 2012
Accepted 4 January 2012
Available online 2 February 2012
Keywords:
Ships propeller jet
Efux velocity
Seabed scour

a b s t r a c t
The present study proposed the semi-empirical methods for determining the efux velocity from a ships
propeller. Ryan [1] dened the efux velocity as the maximum velocity taken from a time-averaged
velocity distribution along the initial propeller plane. The Laser Doppler Anemometry (LDA) and Computational Fluid Dynamics (CFD) were used to acquire the efux velocity from the two propellers with
different geometrical characteristics. The LDA and CFD results were compared in order to investigate
the equation derived from the axial momentum theory. The study conrmed the validation of the axial
momentum theory and its linear relationship between the efux velocity and the multiplication of the
rotational speed, propeller diameter and the square root of thrust coefcient. The linear relationship of
these two terms is connected by an efux coefcient and the value of this efux coefcient reduced when
the blade number increased.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
The investigations into predicting the velocity within the ships
propeller jet which can lead to seabed scouring are of particular
interest for the design of marine structures. In Whitehouses [2]
book Scour at Marine Structures, the potential damage made by
the propeller jet was highlighted. The action of the propeller jet
to the seabed scouring was also described in Sumer and Fredse
[3] book The Mechanics of Scour in the Marine Environment and
Gaythwaite [4] book Design of Marine Facilities for the Berthing,
Mooring, and Repair of Vessels. The jet impingement of a ro-ro
ship to the seabed is illustrated in Fig. 1.
The parameters to investigate the seabed scouring are presented in Fig. 2. The velocity eld of a ships propeller jet was
preliminary investigated in order to determine the propeller
induced seabed scouring. The inuences of the rudder [1,5], hull
and the berth geometry to the velocity within the jet would
normally be included after the unconned propeller jet being
established. The impingement velocities were therefore used to
determine the erosion extent and erosion rate within the seabed
[69]. The inuence of the bed material, which resisted jet impingement, was considered in order to propose an effective remedial

Corresponding author at: Department of Civil Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia.
Tel.: +6 03 7967 7675/44 028 9097 4006; fax: +6 03 7967 5318/44 028 9097 4278.
E-mail addresses: wlam@um.edu.my, joshuawhlam@hotmail.com (W. Lam).
1
Tel.: +44 028 9097 4006; fax: +44 028 9097 4006.
0141-1187/$ see front matter 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apor.2012.01.002

action [10]. The problem of the ships induced scour was also investigated by using a simplied round jet, which has been documented
by Yeh et al. [11] and Ylsel et al. [12]. Lam et al. [13] attempted
to establish the velocity prediction at the efux plane from a ships
propeller. Isbash [14] and PIANC [15] provided procedures for estimating the contributions of numerous factors that determined the
seabed damage.
Previous researchers used the maximum velocity at the initial
stage as an input to investigate the ships propeller jet induced
scour. The velocity distribution within the entire propeller jet, the
bed velocity with or without rudder and the inuence of the jet to
the sediment deposition were later predicted using these available
equations [1622]. The accurate prediction of the efux velocity
became important since it was an initial input in the entire predicting system.
The objectives of this paper are to investigate: (1) the validity of
the axial momentum theory used to predict the efux velocity; (2)
linear relationship between the efux velocity and the multiplication of rotational speed, propeller diameter; (3) inuences of the
propeller geometry to the efux velocity prediction; (4) proposal
of efux coefcients from different propellers. This paper presents
the objectives of the current research initially and then followed
by presenting the methodology used. The validity of the LDA and
CFD results is discussed in Section 3. Section 4 is the main part of
this paper with ve sub-sections. The rst subsection describes the
axial momentum theory and the second subsection describes the
limitation of theory. The third subsection describes the previous
works of the axial momentum equation by considering the geometrical characteristics. The fourth and fth subsections discuss

W. Lam et al. / Applied Ocean Research 35 (2012) 1424

15

Fig. 1. Jet impingement in the case of ro-ro ship with stern ramp [3].

the LDA and CFD results in terms of the efux velocity and efux
coefcient.

Fig. 3. Schematic of experimental tank layout (plan view).


Table 1
Propeller characteristics.

2. Methodology
A joint experimental and numerical approach was carried out
in order to acquire the time-averaged velocity within a rotating
ships propeller jet. Experimental measurements provided direct
data within the diffusing jet and the numerical modelling gave data
at high rotational speeds that was hard to achieve in experiments.
The numerical modelling was able to provide additional blades to
the virtual model in order to investigate the blade inuences to the
velocity.
2.1. Experimental measurements
The experiments are carried out in order to acquire the mean
velocity within a diffusing jet. A water tank which was large enough
to allow the unhindered expansion and diffusion of the propeller jet
was used in a series of experiments. The purpose-built test tank was
7.5 m 4.4 m in plan by 1.0 m deep, as shown in Fig. 3. The water
tank was equipped with a drainage system from which the water
could be drained out for cleaning after the experiment ended. An
overow spillway was also designed in order to prevent ooding.
The tank was lled to a height of 880 mm in 1 h via a 50 mm diameter water pipe. The propeller shaft was located at almost mid-depth
in the tank. The effect of the tank bottom and water boundaries
was not found to inuence the free expansion of the unconned jet
under investigation.
The present study was undertaken using two propellers at bollard pull condition (zero advance speed), as shown in Table 1 and
Fig. 4. Two propellers are termed as propeller-76 and propeller-131

Fig. 2. Parameters of the seabed scouring [10].

Propeller diameter, Dp
Hub diameter, Dh
Blade number, N
Rake angle, 
Blade area ratio,
Thrust coefcient, Ct
Mean pitch ratio (P )

Propeller-76

Propeller-131

76 mm
15.2 mm
3
0
0.473
0.4
1

131 mm
35.0 mm
6
0
0.922
0.56
1.14

according to their diameters in millimetre. The propeller was tted


to a stainless steel shaft on a stationary rig to allow the rotation
at the zero advance speeds. A central nut was used to lock the
propeller to the shaft. The propeller rig was 0.47 m 0.57 m in
plan by 1.57 m high and was fabricated using the galvanised steel
of 50 mm 50 mm in section by 5 mm in thickness. The propeller
rig was located along the centreline of the water tank about 1.5 m
from the rear. The circulation effect of water within the tank has
been shown to be insignicant to the jet expansion [20,21].
A direct current electric motor was employed to rotate the
propellers at constant speeds. The motor was made by GEC Electromotors Limited. It was rated at 0.55 kW and the design speed is
800 rpm. The design speed is the speed for maximum efciency of
a motor. This motor used in this experiment has performed well by
gearing the output to a desire speed. In this experiment, the motor
was tted at the upper part of the rig at a safe distance of 0.5 m

Fig. 4. Geometry of a six-bladed propeller and a three-bladed propeller: (a) aft view
and (b) side view.

16

W. Lam et al. / Applied Ocean Research 35 (2012) 1424

Fig. 5. Front view and side view of the measurement grid.

above water surface. The rotational speed was adjusted using the
speed meter which was attached at the wall. Once the motor was
switched on, the motor transferred the torque force to rotate the
propeller.
Laser Doppler Anemometry (LDA) measurements were undertaken to acquire data at the initial plane of a ships propeller jet
for efux velocity study and CFD validation. LDA is a non-intrusive
optical technique used to measure time-averaged axial, tangential
and radial components of velocity along with turbulence levels.
In this experiment laser beams from the probe were focused onto
the measurement volume where the beams intersected. The measurement volume was a few millimetres long, and the beams
intersection region was termed as the fringe. The Doppler frequency was calculated by the analysis of light scattered from seeded
particles. Flow velocity was calculated from the Doppler frequency
and the fringe distance [23].
The seeding particles were aluminium passivated wet powder
(E3064AR) from Silberline Ltd. The aluminium powder was silver
coloured with a 36 m particle size and density of 1.46 g/cm3 . The
high number of recorded particle showed the particles followed the
uid well. The particle was able to scatter light as a burst to LDA
system. The average recorded burst count due to seeding was 6898
particles with a maximum 14,939 particles and a minimum 3982
particles for 40 s. The measurement at each point was taken for 40 s
to obtain an accurate time-averaged velocity.
2.1.1. Measurement grid
In the current tests, one vertical and one horizontal line were
considered in order to measure the three components of velocity.
The vertical line was able to measure the axial and radial components of velocity. The horizontal line was able to measure the axial
and tangential components of velocity. A 5 mm measurement step
was chosen to measure the velocity eld, which were 0.13Rp for
propeller-76 and 0.08Rp for propeller-131. The measurement grid
at the initial plane is illustrated in Fig. 5. The current measurement
step was two-time ner compared to the 10 mm measurement
step chosen by Hamill [20]. The use of same spatial sampling of
the velocity eld was recommended in the future study, which
may have a better choice for comparison. The ideal location of the
measurement was the plane immediate downstream after the propeller. However, a 10 mm gap between the propeller face and the
measurement point existed as a space for the intersection of laser
beams forming a measurement volume.
2.2. Numerical simulations
Fluent CFD code was used to predict the velocity eld from a
ships propeller rotating at various speeds [24,25,32]. Seil et al. [33],

WS-Atkins-Consultants et al. [34] and Dargahi [8] carried out CFD


investigations on the ow eld from a ships propeller.
Seil et al. [33] from Rolls-Royle investigated the ow eld around
the propeller in order to increase the efciency of a ships propeller. Seil et al. [33] investigated the problem using Fluent CFD
code by constructing a structured mesh with hexahedral cells. Seil
et al. [33] believed that the structured mesh would be able to control the aspect ratio of the downstream adjacent cells effectively.
Despite the unstructured mesh being able to save time on the grid
generation and cope with problems associated with the complex
geometry better, the structured mesh was still considered the better method to predict the ow eld around the propeller. Seil et al.
[33] implemented the RNG k turbulence model derived from the
Boussinesqs approaches in this simulation due to its swirl modication.
WS-Atkins-Consultants et al. [34] provided guidelines to simulate the hydrodynamic performance of a ships propeller. This
example demonstrates the manner in which the predictions for the
ow around a marine propeller can be achieved. The commercial
code CFX-TASCow was used in this simulation. A structured mesh
with hexahedral grid was used in the model due to the restriction of the CFD package. The standard k turbulence model was
the choice in this case and the simulation was performed with the
second order discretisation scheme.
In Dargahis [8] research, the three-dimensional computational
model was generated by using a structured grid. The structured
grid comprised hexahedral elements. Dargahis [8] work has been
solved using both the standard k and the RNG k turbulence
models. The standard k turbulence model is a robust turbulence
model, whereas RNG k turbulence model was the enhanced k
model to predict the swirling ow.
In this investigation, the CFD simulation has been used in order
to predict the velocity in a wider range of rotational speeds. The
current numerical simulation is to obtain the efux velocity at the
rotational speeds in which the measurement is difcult to be taken.
The further documentation of the grid generation and mathematical model will be presented in the following sections.
2.2.1. Grid generation
The helicoidal surfaces of the propeller blades resulted in a
geometry difcult to create. The creation of the propeller geometry
was based on the geometrical data from the technical drawing of
the propellers. By substituting the geometrical data into Eqs. (1)(3)
[26], the points lying at the leading edge and trailing edges can be
obtained.
= tan1

P
2Rp

(1)

sin =

z
L

(2)

cos =

r
L

(3)

where is the pitch angle, P is the pitch distance, Rp is the radius


of propeller, L is the distance of either the trailing edge or leading
edge to the propeller centre depending on the edge point of interest.
The r, , and z are the radial coordinate, angular coordinate and axial
coordinate from the central of propeller, respectively.
An unstructured grid was generated by using Gambit modeller
[36], obeying the Delaunays law, once the propeller domains were
created. The grid independence was tested to ensure that the virtual
model used an optimum grid number. The computational geometries for propeller-76 and propeller-131 are shown in Fig. 6(a) and
(b). In addition, the computational geometry of propeller-76 has
been modied to produce a four-bladed propeller and a ve-bladed
propeller by adding one and two blades, as shown in Fig. 6(c) and

W. Lam et al. / Applied Ocean Research 35 (2012) 1424

17

0.5
76mm x-plane
76mm z-plane

Velocity, Vo (m/s)

0.0

-0.5

-1.0

-1.5

-2.0
-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

1.0

1.5

Lateral section, r/Rp

(d). Simulation tests were undertaken to determine the sensitivity


of the unstructured grid. Three unstructured grids with 77,360 cells,
99,243 cells and 242,121 cells were tested. The grid with 242,121
cells is sufcient to capture the mean velocity within the jet.
2.2.2. Mathematical models
A rotating reference frame approach [32] was used to induce a
rotating propeller jet. The rotating reference frame is less computational consumption compared to the moving mesh method [32].
The standard k turbulence model [37] was used in this simulation. The standard k model is designed to incorporate the
modication of the low-Reynolds number effects and the spreading of the shear ow in predicting free shear ow. Fluent tutorial
used a standard k model to predict the propeller jet from a naval
architects view [35].
The simulations were initially solved by using a rst order
scheme in order to produce a preliminary result, and the second order scheme was performed later to obtain a more accurate
result. In addition, an upwind scheme was used to predict the ow,
in which direction played a signicant role. The SIMPLE scheme
was used to discretise the pressurevelocity coupling [32]. A segregated solution algorithm was used to the Reynolds-Averaging
NavierStokes (RANS) equations [27] sequentially by updating the
property values until the convergence criterion has been satiseds
[32].
A small under relaxation factor [32] was used in this simulation, which non-dimensionlised value of 0.01 for both the pressure
(default 0.3) and momentum (default 0.7) discretisations. After
the solution reached convergence using a rst order discretisation
scheme, the iterative process was continued with a second order
process. The convergence of a second order scheme was achieved
at 34,000 iterations. For the propeller jet simulation, a 2.2 GHz Pentium 4 personal computer with 512 Mb RAM needs 32 h to reach a
converged solution.
3. Validation
The experimental measurements were validated by comparing the velocity prole at various rotational speeds. The velocity
proles of the axial component acquired from the horizontal
and vertical lines were compared to ensure the repeatability and

Velocity, Vo (m/s)

Fig. 6. Computational geometry for CFD simulation: (a) three-bladed propeller; (b)
six-bladed propeller; (c) four-bladed propeller; (d) ve-bladed propeller.

0.5
76mm x-plane
76mm z-plane

0.0

-0.5

-1.0

-1.5

-2.0
-1.5

-1.0

-0.5

0.0

0.5

Lateral section, r/Rp


Fig. 7. Comparison of the measurements from horizontal line (x-plane) and vertical
line (y-plane) from propeller 76 mm at efux plane (10 mm from propeller) with
different rotational speeds: (a) 750 rpm and (b) 1250 rpm.

validity of the experimental results. Hamill [20], Stewart [21],


Hashmi [22] and McGarvey [38] suggested that the velocity
prole of the efux plane was a two-peaked-ridge prole. The
experimental measurements showed the velocity distribution has
a two-peaked-ridge prole for all tested speeds in a range of
7501500 rpm for propeller-76 and 3501000 rpm for propeller131. All the tested results obeyed the widely accepted velocity
prole proposed by the previous researchers.
From the experimental measurements, the region of high velocity gradient showed the average variation between the horizontal
and vertical lines at 750 rpm was only 3%, whereas the rotation axis
showed a maximum variation of 8%, as shown in Fig. 7. The region
of high velocity gradient was the area in between r/Rp = 0.66 and
r/Rp = 0.66. In a propeller jet, the main ow was contributed by the
rotation of blades and the secondary ow was a more complicated
ow with hub and tip vortices. A higher variation at the rotation
axis was expected as the hub vortex occurred at the rotation axis.
The hub vortex was more random in nature. However, the hub
vortex was not part of this study and was therefore treated with
limited concern. The velocity close to the jet boundary was low
and therefore the maximum variation reached 200%.
The horizontal and vertical measurements are not much different as shown in Fig. 7. In a perfect experimental setup, the variation

18

W. Lam et al. / Applied Ocean Research 35 (2012) 1424

4. Results and discussion

0.5
76mm x-plane
76mm z-plane

Velocity, Vo (m/s)

0.0

CFD

-0.5

-1.0

-1.5

-2.0
-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

Lateral section, r/Rp


Fig. 8. Comparison of measurements from horizontal line (x-plane) and vertical line
(y-plane) of propeller-76 at 1000 rpm.

between the horizontal and vertical lines should be equal to zero.


The current variation is mainly due to the minor uctuation of the
rotational speed from the power train which was observed from the
speed meter rather than the performance of the LDA measurement
system.
The validation of CFD results was undertaken by comparing the
LDA measurements with the CFD predictions, as shown in Fig. 8. The
CFD results showed agreement with the LDA measurements with
variation of 13% in the high gradient region in between r/Rp = 0.66
and r/Rp = 0.66. The maximum variation of 65% occurred at the rotation axis. The experimental measurements showed a peak due to
hub effect, but it did not happen in the CFD simulation. The hub vortex was a low-pressure core started from the hub and the strength
of the vortex reduced gradually when moving downstream. Hub
vortex caused the thrust deduction by pulling the propeller boss
and subsequently reduced propeller performance to about 5% [28].
In this research, a low velocity core within the propeller jet was
expected when a coarse grid was designed to capture the timeaveraged velocity with a low computational consumption.
The validation showed that the experimental measurements
acquired from a LDA system have a high repeatability. There is no
much difference for the data acquired using vertical or horizontal lines. The CFD was able to predict the velocity of the propeller
jet except the hub vortex at the rotation axis. The CFD did not
capture the complicated hub vortex that was seen in the experimental measurements. The deciency of the ow calculated from
CFD was limited by a low grid resolution designed to capture the
time-averaged velocity with a low computational consumption.
The current grid resolution was only able to capture the main ow,
but was not capable to capture the hub vortex. All the LDA and CFD
comparison showed a large variation at the rotation axis due to the
deciency of CFD on predicting this highly turbulent region. LDA
showed a peak, but CFD showed a dip at the rotation axis. However, the efux velocity described in this paper was the one with a
distance from the rotation axis, which is 0.67(Rp Rh ) according to
Berger et al. [18]. The variation of efux velocity between LDA and
CFD was in a range of 221% for different propellers. The difculty
of CFD to capture the hub vortex gave insignicant inuences to
the predictions of time-averaged velocity according to the previous
CFD experiences.

After the validation, the maximum velocity at the initial plane is


acquired to investigate the theoretical equation. Efux coefcient
is the dimensionless value obtained by dividing the efux velocity with the rotational speed (rpm), propeller diameter (m) and
square root of thrust coefcient according to the axial momentum
theory. This linear equation was widely accepted by researchers
and the efforts were undertaken to testify the linear relationship of the equation. Some researchers [2022] suggested different
efux coefcients based on the experimental investigation using
limited propeller models. The efux coefcient of 1.59 is widely
accepted due to the strong theoretical support from axial momentum theory. However, the coefcient of 1.59 was derived from a
one-dimensional assumption. The coefcient of 1.59 may be suitable for the plain water jet, but not for a ships propeller jet. The
two-peaked-ridge prole was found at the initial plane of a ships
propeller jet.
4.1. Axial momentum theory
The axial momentum theory is a widely accepted theory used
to predict the efux velocity within a ships propeller jet. Blaauw
and van de Kaa [17], Berger et al. [18], Verhey [19] and Hamill [20]
used the axial momentum theory to describe the characteristics of
a ships propeller jet. The axial momentum theory was proposed
by Froude with reference to Rankines investigations in the 19th
century. The axial momentum theory made six assumptions:
(1) The propeller is represented by an ideal actuator disc of equivalent diameter.
(2) The disc consists of an innite number of rotating blades, rotating at an innite speed.
(3) There is negligible thickness of the disc in the axial direction.
(4) The disc is submerged in an ideal uid (inviscid uid) without
disturbances.
(5) All elements of uid passing through the disc undergo an equal
increase of pressure.
(6) The energy supplied to the disc is, in turn, supplied to the uid
without any rotational effects being induced.
The change of the momentum due to the energy supplied to
the system through the presence of the actuator disc results in a
net thrust on the uid. This thrust can be related to the Bernoullis
equation in order to develop an equation for the efux velocity and
is balanced with the dimensional analysis of the thrust [29]. This
equation can be presented as,
1
2
2
AP (Vdown
Vup
) = Ct n2 Dp4
2

(4)

where  is the density of uid, Ap is the area of actuator disc, Vdown


and Vup are velocity at far downstream and far upstream respectively, Ct is the thrust coefcient of the propeller, n is the speed of
rotation of the propeller in revolutions per second and Dp is the
propeller diameter in metres. The thrust coefcient (Ct ) is a nondimensional coefcient determined from measured performance
characteristics.
As the advance speed (inow velocity to the propeller) VA = 0
and the area Ap = Dp 2 /4, the efux velocity (Vo ), which is the maximum velocity at the face of the propeller [16], becomes the widely
accepted theoretical equation used to predict the efux velocity
Vo = 1.59nDp

Ct

(5)

This equation is the fundamental equation from the axial


momentum theory used to predict the efux velocity within a ships
propeller jet.

W. Lam et al. / Applied Ocean Research 35 (2012) 1424


Table 2
Efux velocity from propeller-76 at horizontal line.

4.2. Limitations of the axial momentum theory


The axial momentum theory only considers the velocity characteristics of the axial component within a submerged jet. A propeller
jet has two other components of velocity (tangential and radial
components) in addition to the axial component of velocity. The
assumption of simulating a rotating propeller jet using a plane submerged jet is therefore not entirely satisfactory, due to the absence
of the tangential and radial components of velocity in a plane
jet [30]. Therefore assumption six in the axial momentum theory
energy supplied . . . to the uid without any rotational effects . . .
is invalid.
Hamill et al. [31] found the assumptions of the axial momentum
theory are not applicable to true propeller jets. Thus, the equations derived from axial momentum theory are inherently awed.
Assumption two in axial momentum theory disc consists of an
innite number of rotating blades, rotating at an innite speed is
clearly void since the true propeller cannot consist of an innite
number of rotating blades and rotate at an innite speed. The propeller is normally carefully designed with three to six blades and the
speed of rotation is carefully chosen to provide the maximum efciency when in service, and tends to be in hundreds of revolutions
per minute.
Assumption three that the disc (and hence propeller) has negligible thickness in the axial direction is not practical as for efcient
operation, the blades of the propeller must have pitch (travel distance of a point in the longitudinal direction of the jet after one
rotation) in this plane. The resulting conclusion that the velocity on either side of the disc (propeller) is approximately the
same is incorrect, with sizable differences by Hamill et al. [31].
As the blades on a ships propeller change in both pitch and area
within a three dimensional space, assumption ve equal increase
of pressure is also invalid as there are signicant differences in
pressure changes across the blade from the root at hub to the
tip.
These shortfalls have led to modication of the theoretical
equations in an attempt to take account of the geometrical characteristics. Hamill et al. [31] found the assumptions in the axial
momentum theory to be inadequate to describe the process
involved in the formation of the propeller jet. Fuehrer et al. [7]
indicated that Eq. (5) was found to be in error as much as 20%.
4.3. Semi-empirical equations for efux velocity
The behaviour of an actual ships propeller jet contradicted most
of the assumptions made in the derivation of the axial momentum
theory. However several researchers, such as Fuehrer and Rmisch
[16], Berger et al. [18], Verhey [19] and Hamill [20] have developed
equations to predict the efux velocity based on the axial momentum theory. The maximum velocity taken from a time-averaged
velocity distribution along the initial propeller plane was termed
the efux velocity denoted as Vo [1].
Hamill [20] rened the theoretical equation of axial momentum
theory for the efux velocity through an experimental investigation of a rotating ships propeller jet instead of a plain water jet.
He proposed a lower coefcient value, producing a semi-empirical
equation based on the detailed measurements on two propellers
Vo = 1.33nDp

Ct

19

(6)

Stewart [21] performed similar experiments on two other propellers and proposed an equation for the efux velocity where the
coefcient was based on geometrical characteristics of the propellers. Stewart [21] reported the coefcient used in the existing
equation to predict efux velocity was not a constant but was

Rotational speed (rpm)

Vo (1)
Vo (2)

750

1000

1250

1500

1.04
1.00

1.34
1.39

1.69
1.72

2.03
2.09

dependent on the propeller characteristics. The following equation


was subsequently suggested:
Vo = nDp

Ct

(7)

where the efux coefcient,  is equal to:


= Dp 0.0686 P 

1.519

0.323

(8)

where is the blade area ratio (projected area of all blades related
to the propeller disc area) and P is the pitch ratio of the propeller
(quotient of a pitch and the propeller diameter). The efux coefcient  in Eq. (8) used a dimensional term (Dp ) in an otherwise
non-dimensional relationship. Hashmi [22] rened this equation
by non-dimensioning the propeller diameter (Dp ), by dividing by
the hub diameter (Dh )
Vo = Eo nDp

Ct

(9)

where the efux coefcient Eo , is equal to:


Eo =

 D 0.403
p

Dh

Ct 1.79 0.744

(10)

4.4. LDA measurements


The time-averaged velocities were measured to acquire the
efux velocity at various rotational speeds for propeller-76 and
propeller-131. The changes of the efux velocity with various
rotational speeds were later plotted in order to obtain the efux
coefcient to compensate the aforementioned weakness of the
axial momentum theory.
4.4.1. Efux velocity from LDA measurement
When the propeller is rotating, the propeller draws the
upstream water into the slipstream and discharges this water
downstream as a high velocity jet. The velocity within the jet
changed when the rotational speeds changed. The velocity distribution immediate downstream of the propeller, which was termed
as efux plane, varied along the propeller blade. The blades contributed to the motion of the jet and therefore a low velocity core
should be expected close to the hub. The secondary ow and the
hub vortex were treated as an additional disturbance to the jet in
this study.
The efux plane has two velocity peaks separated by the rotation axis. Two peaks were measured due to the axis symmetrical
characteristics of the jet as suggested by Hamill [20]. The efux
velocities of propeller-76 were acquired at four different speeds by
using both the horizontal and vertical lines. Two efux velocities
across the horizontal section of propeller-76 are termed as Vo (1)
on the left hand side and Vo (2) on the right hand side (Table 2).
Two efux velocities across the vertical section of propeller-76 are
termed as Vo (3) on the top and Vo (4) at the bottom (Table 3).
For the propeller-131, the same notation system was applied to Vo
(5) for the efux velocity on the left hand side, Vo (6) for the right
hand side, Vo (7) on the top and Vo (8) at the bottom, as shown in
Tables 4 and 5 respectively.
The hydrodynamics of the underwater propeller jet is complicated associated with the swirling features. The hydrodynamic
behaviour at the efux plane is therefore an area of interest for

20

W. Lam et al. / Applied Ocean Research 35 (2012) 1424

Table 3
Efux velocity from propeller-76 at vertical line.
Rotational speed (rpm)

Vo (3)
Vo (4)

750

1000

1250

1500

1.03
1.07

1.33
1.35

1.70
1.72

2.05
2.07

Table 4
Efux velocity from propeller-131 at horizontal line.
Rotational speed (rpm)

Vo (5)
Vo (6)

350

500

750

1000

0.91
0.91

1.25
1.29

1.94
1.99

2.53
2.59

sensitive to the rotational speeds and their locations of data acquisition. The efux coefcients were the values of 1.6901 on the left
acquisition, 1.7244 on the right acquisition, 1.6959 on the top acquisition and 1.7203 at the bottom acquisition even the same propeller
used in the measurements. The sensitivity of the rotational speed to
the efux coefcient was inherited when applying to a larger propeller. The efux coefcients of the propeller-131 were in a range of
1.55891.6565. The average efux coefcients for the propeller-76
and propeller-131 were the values of 1.7077 and 1.6135 respectively. The high efux coefcient of the propeller-76 may be due
to the higher rotational speed rather than the impacts from the
propeller geometry. A more exible study by using the CFD modelling was therefore implemented for a wider range of rotational
speeds.
4.5. CFD predictions
As described in Section 2, the CFD models were created to obtain
the efux velocities and subsequently the efux coefcients from
a wider range of rotational speeds for further investigation.

Table 5
Efux velocity from propeller-131 at vertical line.
Rotational speed (rpm)

Vo (7)
Vo (8)

350

500

750

1000

0.93
0.92

1.32
1.32

2.05
1.98

2.67
2.77

hydrodynamist. The efux velocity increased from the high rotational speeds of 750 rpm up to 1500 rpm. The outcome surprisingly
showed that the rotational speed was the only variable inuencing the magnitude of the efux velocity.
The efux velocity can be
predicted by multiplying the terms of Dp Ct and efux coefcient.
This relationship was further validated by using the results acquired
from the horizontal and vertical lines for propeller-76. The linear
relationship between the efux velocity and the rotational speed
was also validated by using a larger propeller-131 at the lower
rotational speeds, as shown in Tables 25. The study found that
the propeller diameter and thrust coefcient are two xed numbers from the propeller design of naval architect which cannot be
changed. The efux coefcient is the critical constant other than
the rotational speed.
4.4.2. Efux coefcient from LDA measurements
The efux velocities were acquired at various rotational speeds
to investigate the efux coefcient proposals by the 
previous
researchers. The graph of efux velocity Vo versus nDp Ct was
plotted to investigate the coefcient of 1.59 derived from the axial
momentum theory. The LDA measurements of set Vo (1)Vo (8)
are plotted as shown in Table 6 and Fig. 9. The linear lines of all
the sets were plotted to obtain the coefcients from different propellers.
The efux coefcients of propeller-76 were in a range of
1.69011.7244. The study found that the efux coefcient is
Table 6
Efux coefcient from LDA measurements.
Set of
measurement

Propeller

Efux
coefcient

R2

Vo
Vo
Vo
Vo
Vo
Vo
Vo
Vo

Propeller-76
Propeller-76
Propeller-76
Propeller-76
Propeller-131
Propeller-131
Propeller-131
Propeller-131

1.6901
1.7244
1.6959
1.7203
1.5589
1.5961
1.6426
1.6565

0.9985
0.9973
0.9982
0.9963
0.9987
0.9990
0.9988
0.9957

(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)

4.5.1. Efux velocity from CFD results


The CFD models are able to provide a cheaper solution with
less time consuming. The CFD investigated the efux velocity in
a wider range of rotational speeds (250 rpm and 1750 rpm with a
step of 250 rpm), as shown in Table 7. The efux velocities from
three additional rotational speeds were investigated in order to
provide a wider range of data. The comparison of CFD and LDA
at 750 rpm, 1000 rpm, 1250 rpm and 1500 rpm was 0%, 3%, 9% and
19% respectively. The CFD predictions showed high accuracy for
the predictions of the low rotational speeds and their accuracy
decreased with the increase of the rotational speeds. The jet at
the high rotational speed is more complicated and therefore the
CFD may not be able to capture the efux velocity as accurate
as those from the low rotational speed by using a personal computer.
The efux velocities of propeller-76 were predicted at various
rotational speeds by using the axial momentum theory, as shown
in Table 7. The CFD predictions have a closer agreement with the
axial momentum theory at a low rotational speed rather than a
high rotational speed as LDA. A more complicated ow occurred
at a high rotational speed and this ow is more difcult to predict
using CFD. For propeller-131, seven simulations were carried out
at various rotational speeds, as shown in Table 8.
The CFD prediction provided more data to investigate the efux
velocity at the lower and higher ranges of rotational speeds. From
the CFD predictions, the efux velocity remained sensitive to the
rotational speeds regardless the change of rotational speeds. The
efux coefcient is further investigated in the next section.
4.5.2. Efux coefcient from CFD results
The efux coefcients obtained from various data sets are
denoted as Vo (i)Vo (viii), as shown in Table 9. For propeller-76,
both the LDA measurements (points from horizontal and vertical
lines) and CFD
 predictions formed the linear relationship between
Vo and nDp Ct with a coefcient of 1.7077 and 1.4598 respectively,
as shown in Fig. 10(a). The correlation coefcients (R2 ) for
the efux
velocity at various rotational speeds between Vo and nDp Ct were
high for both LDA measurement and CFD prediction, as shown in
Table 9. High correlation meant the high accuracy of the linear relationship. For propeller-131, both the LDA measurements and CFD
showed also the linear relationship between Vo and
predictions

nDp Ct , as shown in Fig. 10(b). The CFD coefcient was lower than
the LDA coefcient since the predicted CFD values were low.

W. Lam et al. / Applied Ocean Research 35 (2012) 1424

2.5

2.5

y = 1.6901x
R2 = 0.9985

2.0

1.5

Vo (m/s)

Vo (m/s)

y = 1.7244x
R2 = 0.9973

2.0

1.5

1.0

0.5

1.0

0.5

0.0

0.0
0.5

0.0

1.0

1.5

0.5

0.0

n Dp Ct

d
y = 1.6959x
R2 = 0.9982

2.0

Vo (m/s)

Vo (m/s)

y = 1.7203x
R2 = 0.9963

1.5

1.0

0.5

1.0

0.5

0.0

0.0
0.0

0.5

1.0

0.0

1.5

0.5

n Dp Ct

f
y = 1.5589x
R2 = 0.9987

2.5

1.5

3.0

y = 1.5961x
R2 = 0.999

2.5
2.0

Vo (m/s)

2.0

Vo (m/s)

1.0

n Dp Ct

3.0

1.5
1.0

1.5
1.0
0.5

0.5

0.0

0.0
0.0

0.5

1.0

1.5

0.0

2.0

0.5

n Dp Ct

1.0

1.5

2.0

1.5

2.0

n Dp Ct

3.0

3.0

y = 1.6426x
R2 = 0.9988

2.5

y = 1.6565x
R2 = 0.9957

2.5

2.0

2.0

Vo (m/s)

Vo (m/s)

1.5

2.5

2.0

1.5

1.0

n Dp Ct

2.5

21

1.5
1.0
0.5

1.5
1.0
0.5

0.0

0.0

0.0

0.5

1.0

1.5

n Dp Ct
Fig. 9. Relationship of Vo and nDp

2.0

0.0

0.5

1.0

n Dp Ct

Ct from various measurement sets Vo (1)Vo (8) corresponding to (a)(h).

22

W. Lam et al. / Applied Ocean Research 35 (2012) 1424

Table 7
Efux velocity from axial momentum theory and CFD for propeller-76.
Rotational speed (rpm)

Axial momentum theory (m/s)


CFD (m/s)
Variation

250

500

750

1000

1250

1500

1750

0.32
0.31
3%

0.64
0.65
2%

0.96
1.04
9%

1.27
1.30
2%

1.59
1.53
4%

1.91
1.64
14%

2.23
1.93
13%

Table 8
Efux velocity from axial momentum theory and CFD for propeller-131.
Rotational speed (rpm)

Axial momentum theory (m/s)


CFD (m/s)
Variation

250

350

500

750

1000

1250

1500

0.65
0.58
11%

0.91
0.79
13%

1.30
1.03
21%

1.95
1.72
12%

2.60
2.31
11%

3.25
2.89
11%

3.90
3.46
11%

Table 9
Efux coefcient from axial momentum theory, LDA measurements and CFD predictions.
Set

Source of data

Propeller

Efux coefcient

R2

Vo
Vo
Vo
Vo
Vo
Vo
Vo
Vo
Vo

LDA
CFD
LDA
CFD
Axial momentum theory
LDA
CFD
CFD
CFD

Propeller-76
Propeller-76
Propeller-131
Propeller-131
Actuator disc
Both propeller-76 and propeller-131
Both propeller-76 and propeller-131
Modied four-bladed propeller-76
Modied ve-bladed propeller-76

1.7077
1.4598
1.6135
1.4051
1.5900
1.6512
1.4092
1.4812
0.8948

0.9963
0.9596
0.9936
0.9980
1.0000
0.9868
0.9790
0.9820
0.9287

(i)
(ii)
(iii)
(iv)
(v)
(vi)
(vii)
(viii)
(ix)

The efux coefcient from the axial momentum theory without consideration of the geometrical characteristics was compared
to the LDA and CFD coefcients. The efux coefcient was
plotted by using all the LDA efux velocities at various rotational speeds regardless the propellers used. The efux coefcient
of CFD results was also plotted by using all the CFD predictions at various rotational speeds. The efux coefcient obtained
from the LDA measurements regardless the propeller geometry was 1.6512, whereas the efux coefcient of CFD results
regardless the propeller geometry was 1.4092, as shown in
Fig. 10(c) and Table 9. The efux coefcient of 1.59 from the
axial momentum theory fell in between the LDA coefcient and
the CFD coefcient. The axial momentum theory has assumed
an actuator disc with innite blade number. The results showed
that the efux coefcient reduced when the blade number
increased.
4.5.3. Inuences of the propeller geometry to efux velocity
The geometrical characteristics of propeller-76 and propeller131 were different in terms of propeller diameter, blade number,
blade area ratio, thrust coefcient and mean pitch ratio. Inuences of the propeller geometry to the velocity eld within a
ships propeller jet remained an interesting unknown in this area.
LDA measurements showed the efux coefcient reduced when
the blade number increased. The efux coefcients of actuator
disc, six-bladed propeller and three-bladed propeller increased
from 1.59 to 1.6135 and then to 1.7077, as shown in Table 9.
The blade number may give signicant inuence to the values of
efux coefcient. The CFD was therefore implemented to investigate the inuences of blade number to the efux coefcient
by adding one and two additional blades on the three-bladed
propeller-76.

Stewart [21] and Hashmi [22] conducted studies to investigate


the inuence of the propeller geometry to the efux coefcients.
The three-bladed propeller has a LDA efux coefcient of 1.71,
whereas the six-bladed propeller and the actuator disc of axial
momentum theory have efux coefcients of 1.61 and 1.59 respectively. The six-bladed propeller was closer to the actuator disc
condition compared to the three-bladed propeller. The efux
coefcients for the four-bladed and ve-bladed propellers were
predicted as 1.4812 and 0.8948 by using the CFD model, as shown
in Fig. 11.
The CFD results of the four-bladed and ve-bladed virtual propellers gave important references of the blade number inuences
to the efux coefcient. The efux coefcient of the four-bladed
propeller was 1% higher than that of the three-bladed propeller,
whereas the efux coefcient of the ve-bladed propeller was 39%
lower than that of the three-bladed propeller. The geometrical

Table 10
Proposed equations based on propeller with different geometrical characteristics.
Propeller

Source of
data

Propeller-76

LDA

Propeller-76

CFD

Propeller-131

LDA

Propeller-131

CFD

Both propeller

LDA

Both propeller

CFD

Modied four-bladed propeller-76

CFD

Modied ve-bladed propeller-76

CFD

Proposed equation


Ct
Vo = 1.46nDp
Ct
Vo = 1.61nDp
Ct
Vo = 1.41nDp
Ct
Vo = 1.65nDp
Ct
Vo = 1.41nDp
Ct
Vo = 1.48nDp
Ct
Vo = 1.71nDp

Vo = 0.89nDp

Ct

W. Lam et al. / Applied Ocean Research 35 (2012) 1424

23

4.0

4.5
4.0

y = 1.4812x
R2 = 0.982

3.5

3.5

2.5
2.0
1.5

LDA
CFD

1.0

y = 0.8948x
R2 = 0.9287

2.5

y = 1.4598x
R2 = 0.9596

Vo (m/s)

Vo (m/s)

3.0

y = 1.7077x
R2 = 0.9963

3.0

2.0
1.5
four-bladed
five-bladed

1.0

0.5
0.0
0.0

1.0

2.0

3.0

n Dp Ct

0.5
0.0
0.0

4.5

3.0

Fig. 11. Efux coefcient from CFD prediction by using two virtual propellers.

3.5
y = 1.6135x
R2 = 0.9936

3.0

Vo (m/s)

2.0

n Dp Ct

4.0

5. Conclusions

2.5

The applications of LDA and CFD to improve the efux velocity


prediction have been demonstrated. The semi-empirical equations
are proposed in Table 10 and the ndings are as follows:

y = 1.4051x
R2 = 0.998

2.0
1.5

LDA
CFD

1.0
0.5
0.0
0.0

1.0

2.0

3.0

n Dp Ct

1.0

4.5

(1) LDA and CFD results conrmed the validity of the axial momentum theory to predict the efux velocity from
 a ships propeller
by relating the efux velocity, Vo with nDp Ct linearly.
(2) LDA and CFD results conrmed the linear line intercepting at
zero position with a strong correlation coefcient.
(3) LDA and CFD results showed the efux velocity increased with
the rotational speeds.
(4) The LDA measurement showed that the efux coefcient
reduced when the blade number increased.

4.0

Acknowledgements

3.5

y= 1.6512x
R2 = 0.9868

Vo (m/s)

3.0
2.5

The current research was supported by SPUR studentship from


Queens University Belfast and National Natural Science Foundation
of China (grant no. 51006019).

y = 1.4092x
R2 = 0.979

2.0

References

1.5
Axial Momentum Theory

1.0

LDA

0.5

CFD

0.0
0.0

1.0

2.0

3.0

n Dp Ct
Fig. 10. Efux coefcient of axial momentum theory, LDA measurement and CFD
prediction from data points of (a) propeller-76; (b) propeller-131; (c) both propeller76 and propeller-131.

characteristics of propeller-76 were designed with an optimum


blade number of three. The off-design condition of the four-bladed
and ve-bladed propellers may inuence the performance of the
propeller and subsequently the efux coefcient. Blade area ratio
(AE /A0 ) becomes larger with a large number of blade areas. The
blade area becomes linearly larger with the number of blades.

[1] Ryan D. Methods for determining propeller wash induced scour in harbours.
Ph.D. thesis. Thesis submitted to the Queens University of Belfast for the degree
of Doctor of Philosophy; 2002.
[2] Whitehouse R. Scour at marine structures: a manual for practical applications.
Thomas Telford; 1998.
[3] Sumer BM, Fredse J. The mechanics of scour in the marine environment. World
Scientic; 2002.
[4] Gaythwaite J. Design of marine facilities for the berthing, mooring, and repair
of vessels. ASCE Publications; 2004.
[5] Hamill GA, Ryan D, Johnston HT. Effect of rudder angle on propeller wash velocities at a seabed. Proceedings of the Institution of Civil Engineers: Maritime
Engineering 2009;162:2733.
[6] Bergh H, Cederwall K. Propeller erosion in harbours. Bulletin No. TRITA-VBI107. Stockholm, Sweden: Hydraulics Laboratory, Royal Institute of Technology;
1981.
[7] Fuehrer M, Rmisch K. Propeller jet erosion and stability criteria for bottom
protection of various constructions. PIANC Bulletin 1987;(58).
[8] Dargahi B. Three-dimensional modelling of ship-induced ow and erosion. Proceeding of the Institution of Civil Engineers: Water & Maritime Engineering
2003;(2):193220.
[9] van Blaaderen EA. Modelling bow-thrusters induced ow near a quay wall.
Masters thesis. Delft: University of Technology; 2006.
[10] Prosser M. Propeller induced scour. Tech. rep., BHRA Project RP A01415. Craneld: The Fluid Engineering Centre; 1986.

24

W. Lam et al. / Applied Ocean Research 35 (2012) 1424

[11] Yeh PH, Chang KA, Henriksen J, Edge B, Chang P, Silver A, et al. Large-scale
laboratory experiment on erosion of sand beds by moving circular vertical jets.
Ocean Engineering 2009;36:24855.
[12] Ylsel A, Celikoglu Y, Cevik E, Yksel Y. Jet scour around vertical piles and pile
groups. Ocean Engineering 2005;32:34962.
[13] Lam W, Hamill GA, Robinson DJ, Raghunathan S. Observations of the initial 3D
ow from a ships propeller. Ocean Engineering 2010;37:13808.
[14] Isbash S. Construction of dams by dumping stone in running water. MoscowLeningrad; 1935 [see also, Hydraulics of river channel closure. London:
Butterworths; 1970].
[15] PIANC. Review of selected standards for oating dock designs. Tech. rep., Special
report of the SPN Commission January 1997 issue, Supplement to Bulletin nr.
93, RecCom Working Group special 1997; 1997.
[16] Fuehrer M, Rmisch K. Effects of modern ship trafc on islands and ocean
waterways and their structures. In: P.I.A.N.C. 24th congress, Section 13. 1977.
[17] Blaauw HG, van de Kaa EJ. Erosion of bottom and sloping banks caused by the
screw race of manoeuvering ships. Netherlands: Delft Hydraulics Laboratory;
July 1978. Publication No. 202.
[18] Berger W, FelKel K, Hager M, Oebius H, Schale E. Courant provoque par les
bateaux protection des berges et solution pour eviter lerosion du lit du haut
rhin. In: P.I.A.N.C. 25th congress, Section I-1. 1981.
[19] Verhey HJ. The stability of bottom and banks subjected to velocities in the propeller jet behind ships. Netherlands: Delft Hydraulics Laboratory; April 1983.
Delft Publication No. 303.
[20] Hamill GA. Characteristics of the screw wash of a manoeuvering ship and the
resulting bed scour. Ph.D. thesis. Thesis submitted to the Queens University of
Belfast for the degree of Doctor of Philosophy; 1987.
[21] Stewart DPJ. Characteristics of a ships screw wash and the inuence of quay wall
proximity. Ph.D. thesis. Thesis submitted to the Queens University of Belfast
for the degree of Doctor of Philosophy; 1992.
[22] Hashmi HN. Erosion of a granular bed at a quay wall by a ships screw wash.
Ph.D. thesis. Thesis submitted to the Queens University of Belfast for the degree
of Doctor of Philosophy; 1993.
[23] Dantec. BSA ow software version 2.1 users guide and installation. Skovlunde:
Dantec Dynamics; 2003.

[24] Versteegs HK, Malalasekera W. An introduction to computational uid dynamics the nite volume method. Essex: Prentice Hall; 1995.
[25] Donea J, Huerta A. Finite element methods for ow problems. John Wiley and
Sons; 2002.
[26] Carlton JS. Marine propellers and propulsion. London: Butterworth Heinemann; 1994.
[27] Chen CJ, Jaw SY. Fundamentals of turbulence modelling. Taylor and Francis
Publisher; 1998.
[28] Ochi F, Fujisawa T, Ohmori T, Kawamura T. Simulation of propeller hub vortex
ow. In: First international symposium on marine propulsors SMP09. 2009.
[29] Lam W, Hamill GA, Song YC, Robinson DJ, Raghunathan S. A review of the
equations used to predict the velocity distribution within a ships propeller
jet. Ocean Engineering 2011;38(January (1)):110.
[30] Qurrain R. Inuence of the sea bed geometry and berth geometry on the hydrodynamics of the wash from a ships propeller. Ph.D. thesis. Thesis submitted
to the Queens University of Belfast for the degree of Doctor of Philosophy;
1994.
[31] Hamill GA, McGarvey JA, Hughes DAB. Determination of the efux velocity from
a ships propeller. Proceedings of the Institution of Civil Engineers: Maritime
Engineering 2004;157(2):8391.
[32] Fluent User Manual. Fluent Users Guide. Lebanon, USA: Fluent Inc; 2003.
[33] Seil GJ, Lunberg J, Peterson G. CFD Calculation and Experimental Validation of
a Kamewa High-skew Marine Propeller. Proceeding of CFD 2003: Computational uid dynamics technology in Ship Hydrodynamics. London, pp 137-148;
2003.
[34] WS Atkins Consultants et al. Best Practice Guidelines for Marine Applications
of Computational Fluid Dynamics, 2004.
[35] Propeller Flow Field Simulation using MRF Model. Fluent Tutorial. New Hampshire: Fluent Inc; 2004.
[36] Gambit. Gambit Modelling Guide. Lebanon, USA: Fluent Inc; 2003.
[37] Wilcox DC. Turbulence Modelling for CFD. La Canada, California: DCW Industries, Inc; 1998.
[38] McGarvey JA. The Inuence of the Rudder on the Hydrodynamics and the
Resulting Bed Scour of a Ships Screw Wash. PhD Thesis. Thesis submitted to
the Queens University of Belfast for the degree of Doctor of Philosophy; 1996.

You might also like